RT info:eu-repo/semantics/article
T1 Automated multiclass classification of spontaneous EEG activity in Alzheimer’s disease and mild cognitive impairment
A1 Ruiz Gómez, Saúl José
A1 Gómez Peña, Carlos
A1 Poza Crespo, Jesús
A1 Gutiérrez Tobal, Gonzalo César
A1 Tola Arribas, Miguel Ángel
A1 Cano, Mónica
A1 Hornero Sánchez, Roberto
K1 Alzheimer’s disease
K1 Mild cognitive impairment
K1 Electroencephalography (EEG)
K1 Spectral analysis
K1 Nonlinear analysis
K1 Multiclass classification approach
K1 12 Matemáticas
K1 32 Ciencias Médicas
AB The discrimination of early Alzheimer’s disease (AD) and its prodromal form (i.e., mildcognitive impairment, MCI) from cognitively healthy control (HC) subjects is crucial since thetreatment is more effective in the first stages of the dementia. The aim of our study is to evaluate theusefulness of a methodology based on electroencephalography (EEG) to detect AD and MCI. EEGrhythms were recorded from 37 AD patients, 37 MCI subjects and 37 HC subjects. Artifact-free trialswere analyzed by means of several spectral and nonlinear features: relative power in the conventionalfrequency bands, median frequency, individual alpha frequency, spectral entropy, Lempel–Zivcomplexity, central tendency measure, sample entropy, fuzzy entropy, and auto-mutual information.Relevance and redundancy analyses were also conducted through the fast correlation-based filter(FCBF) to derive an optimal set of them. The selected features were used to train three differentmodels aimed at classifying the trials: linear discriminant analysis (LDA), quadratic discriminantanalysis (QDA) and multi-layer perceptron artificial neural network (MLP). Afterwards, each subjectwas automatically allocated in a particular group by applying a trial-based majority vote procedure.After feature extraction, the FCBF method selected the optimal set of features: individual alphafrequency, relative power at delta frequency band, and sample entropy. Using the aforementionedset of features, MLP showed the highest diagnostic performance in determining whether a subject isnot healthy (sensitivity of 82.35% and positive predictive value of 84.85% for HC vs. all classificationtask) and whether a subject does not suffer from AD (specificity of 79.41% and negative predictivevalue of 84.38% for AD vs. all comparison). Our findings suggest that our methodology can helpphysicians to discriminate AD, MCI and HC.
PB MDPI
YR 2018
FD 2018
LK https://uvadoc.uva.es/handle/10324/57386
UL https://uvadoc.uva.es/handle/10324/57386
LA eng
NO Entropy, 2018, vol. 20, n. 1, p. 35
NO Producción Científica
DS UVaDOC
RD 22-dic-2024